TWI833276B - Sleep-wake determination system, sleep-wake determination method, and program - Google Patents

Abstract

To provide a sleep-wake determination system or the like for determining sleep and wake with sufficiently high accuracy using a small number of wearing devices. According to an aspect of the present invention, a sleep-wake determination system is provided. The sleep-wake determination system comprises at least one processor configured to execute a program in such a manner that each of the following steps can be performed. In an acquisition step, a signal indicating acceleration of at least a part of a body of a user is acquired. In a band limitation step, a frequency component included in the signal indicating acceleration is limited to specific frequency component. In a transformation step, the signal configured of the specific frequency component is Fourier transformed to generate data for determination. In a determination step, sleep and wake of the user is determined based on the data for determination and preset reference information.

Description

Sleep wakefulness determination system, sleep wakefulness determination method and computer program

The invention relates to a sleep awakening determination system, a sleep awakening determination method and a computer program.

As we all know, if you want to maintain good health, you must ensure healthy sleep, and sleep disorders such as insomnia, sleep apnea, and narcolepsy are harmful to health. To understand someone's sleep status, it is necessary to check the person's actual sleep status over a period of one night to several days.

As a person who examines a person's sleep state, a whole-night sleep polyphasograph (polysomnography; PSG) proposed in Patent Document 1 and the like is being developed. In PSG, multiple electrodes or sensors are installed on the body of the test subject, and the electrodes or sensors are connected to dedicated measuring instruments to measure fundamentals such as brain waves, electrocardiogram, electromyography, and respiratory status. data, and check the status of sleep and wakefulness based on the aforementioned basic data.

[Prior technical literature] [Patent Document]

[Patent Document 1] Japanese Patent Application Publication No. 2013-99507.

In the sleep examination using PSG such as Patent Document 1, many measuring instruments are used, and the examination location is limited to hospitals or laboratories. Therefore, many people cannot easily perform long-term inspections for more than a few days. Also, being in an environment different from your own home, with multiple electrodes or sensors attached to the body, can cause stress and make it difficult to sleep. Therefore, it is difficult to accurately check the usual sleep status.

The present invention has been accomplished in view of the above circumstances, and an object thereof is to provide a sleep-awakening determination system and the like that can determine sleep and awakening with sufficiently high accuracy using fewer wearing devices.

According to an aspect of the present invention, a sleep and awakening determination system is provided. The aforementioned sleep awakening determination system includes at least one processor, and the aforementioned at least one processor can execute a computer program to perform the following steps. In the acquisition step, a signal indicating the acceleration of at least a part of the user's body is acquired. In the band limiting step, the frequency components contained in the signal representing the acceleration are limited to specific frequency components. In the transformation step, a signal consisting of a specific frequency component is Fourier transformed to generate data for determination. In the determination step, the user's sleep and wakefulness are determined based on the determination data and preset reference information.

According to this aspect, sleep and wakefulness can be determined with sufficiently high accuracy using fewer wearing devices.

1: Sleep and wakefulness determination system

2: Wearable devices

2a: Wearable devices

2b: Wearable devices

3:Information processing device

20: Communication bus

21:Communication Department

22:Memory department

23:Control Department

24:Acceleration sensor

30: Communication bus

31: Ministry of Communications

32:Memory department

33:Control Department

331: Acquisition Department

332: Band Restriction Department

333:Conversion Department

334:Correction Department

335:Judgment Department

336:Display control department

A101~A108:Activity

C: Specific frequency component

E:Period

HF: high pass filter

IF: Reference information

IF1: Learned model

M: memory media

f: feature quantity

t: scheduled time

v: three-dimensional acceleration vector

[Fig. 1] Fig. 1 is a diagram showing an outline of the structure of the sleep/awakening determination system 1 according to this embodiment.

[FIG. 2] is a block diagram showing the hardware structure of the sleep/awakening determination system 1 shown in FIG. 1.

[Fig. 3] is a photograph showing an example of the wearable device 2.

[Fig. 4] is a functional block diagram of the control unit 33 of the information processing device 3.

[Fig. 5] is an activity diagram showing the information processing flow of the sleep/awake determination system 1.

[Fig. 6] It is a conceptual diagram for explaining the period (Epoch) E.

[Fig. 7] is a conceptual diagram for explaining period E.

[Fig. 8] A diagram showing the L2 norm of acceleration for each period (Fig. 8A) and a diagram showing a hypnogram (Fig. 8B) showing actual sleep are shown.

[Fig. 9A to Fig. 9D] are graphs showing the accuracy of the learned model IF1 at each sampling frequency.

[Fig. 10] is a graph showing the accuracy of sleep-wake determination at each bit rate in analog-digital conversion.

[Fig. 11] is a schematic diagram showing the standardization process executed by the correction unit 334.

Embodiments of the present invention will be described below using drawings. Various features shown in the embodiments shown below can be combined with each other.

However, a computer program for implementing the software presented in this embodiment may be used as a computer program The read non-transitory computer-readable medium (Non-Transitory Computer-Readable Medium) is provided, or it can be provided by downloading from an external server, or it can be provided by starting the computer program on an external computer and implementing it on the client terminal. Its function (also known as cloud computing).

In addition, in this embodiment, "part" may include, for example, a combination of hardware resources implemented by generalized circuits and software information processing that can be specifically implemented by these hardware resources. In addition, although various information is processed in this embodiment, the information is represented by a physically meaningful value such as a signal value representing a voltage or current, or as a binary bit set composed of 0 or 1. ) The level of the signal value can also be represented by quantum superposition (also known as qubit), and can perform communication or calculations on a generalized circuit.

In addition, a circuit in a broad sense is a circuit realized by appropriately combining at least a circuit (Circuit), a circuit class (Circuitry), a processor (Processor), a memory (Memory), and the like. In other words, it includes application specific integrated circuits (ASICs), programmable logic devices (for example, simple programmable logic devices (SPLD)), complex programmable logic devices (Complex Programmable Logic Devices) ; CPLD) and Field Programmable Gate Array (Field Programmable Gate Array; FPGA)), etc.

1. Hardware structure

In this section, the overall structure of the sleep-wake determination system 1 will be explained.

1.1 Sleep and wakefulness determination system 1

FIG. 1 is a diagram showing an outline of the structure of the sleep/awakening determination system 1 according to this embodiment. The sleep-wake determination system 1 is a system including a wearable device 2 and an information processing device 3. These components can be electronically Sexual communication means exchange of information. FIG. 2 is a block diagram showing the hardware structure of the sleep-wake determination system 1 shown in FIG. 1 .

1.2 Wearable devices 2

FIG. 3 is a photograph showing an example of the wearable device 2 . As shown in FIG. 3 , the wearable device 2 is, for example, a small device that can be worn on the user's arm. In addition, as shown in FIG. 2 , the wearable device 2 has a communication unit 21 , a memory unit 22 , a control unit 23 , and an acceleration sensor 24 . These components are electrically connected inside the wearable device 2 through the communication bus 20 . connection. Each component is further explained below.

The communication part 21 is preferably USB (Universal Serial Bus; Universal Serial Bus), IEEE (Institute of Electrical and Electronics Engineers; American Institute of Electrical and Electronics Engineers) 1394, and Thunderbolt (registered trademark, translated as thunder in Chinese, published by Intel Connector standard), wired LAN (Local Area Network; local area network) network communication and other wired communication means, or may include wireless LAN network communication, 3G (Third Generation Mobile Communication; third generation mobile communication)/LTE as needed (Long Term Evolution; Long Term Evolution Technology)/5G (Fifth Generation Mobile Communication; fifth generation mobile communication) and other mobile communications, Bluetooth (registered trademark) communications, etc. are implemented. Particularly in this embodiment, the communication unit 21 is preferably configured to be able to write information including a time series of three-dimensional acceleration vectors v(x, y, z) measured by the acceleration sensor 24 to be described later, into an external device. Memory media M. The type or form of the memory medium M is not particularly limited. For example, flash memory, card memory, optical disc, etc. can be appropriately used.

The memory unit 22 stores various information defined in the above description. It can be implemented, for example, as a storage device such as a solid state drive (SSD), or as a memory device related to program operations. The temporary necessary information (parameters, arrays, etc.) is implemented in a memory such as Random Access Memory (RAM). In addition, a combination of these can also be used. In particular, information including a time series of three-dimensional acceleration vectors v(x, y, z) measured by the acceleration sensor 24 described below can be memorized. It should be noted that it can also be implemented as directly storing it in the aforementioned storage medium M without using the storage unit 22 .

The control unit 23 performs processing and control of the overall operations related to the wearable device 2 . The control unit 23 is, for example, a central processing unit (Central Processing Unit; CPU) not shown. The control unit 23 realizes various functions related to the wearable device 2 by reading the predetermined computer program stored in the memory unit 22 . That is, the information processing performed by the software stored in the memory unit 22 is specifically implemented by the control unit 23 as an example of hardware, and can be executed as each functional unit included in the control unit 23 . In addition, the control unit 23 is not limited to a single one, and may be implemented with a plurality of control units 23 according to each function. In addition, a combination of these is also possible.

The acceleration sensor 24 is configured to measure the acceleration of a part of the user's body, such as an arm, as three-dimensional vector information. That is, information including a time series of three-dimensional acceleration vectors v(x, y, z) can be obtained from the user.

1.3 Information processing device 3

As shown in FIG. 2 , the information processing device 3 includes a communication unit 31 , a memory unit 32 and a control unit 33 . These components are electrically connected inside the information processing device 3 through the communication bus 30 . Each component is further explained below.

The communication part 31 is preferably USB (Universal Serial Bus; Universal Serial Bus), IEEE (Institute of Electrical and Electronics Engineers; American Institute of Electrical and Electronics Engineers) 1394, and Thunderbolt (registered trademark, translated as thunder in Chinese, published by Intel connector standard), Wired communication means such as wired LAN (Local Area Network; regional network) network communication, or wireless LAN network communication, 3G (Third Generation Mobile Communication; third generation mobile communication)/LTE (Long Term Evolution) as needed; Long-term evolution technology)/5G (Fifth Generation Mobile Communication; fifth generation mobile communication) and other mobile communications, Bluetooth (registered trademark) communications, etc. are implemented. In particular, in this embodiment, the communication unit 31 is preferably implemented as a storage medium reading unit capable of reading information stored in an external storage medium M. In the memory medium M, information including a time series of three-dimensional acceleration vectors v(x, y, z) obtained from the user through the wearable device 2 is stored. Thereby, the communication unit 31 as a storage medium reading unit is configured to be able to read the three-dimensional acceleration vector v(x, y, z) stored in the storage medium M.

The memory unit 32 stores various information defined in the above description. It can be implemented, for example, as a storage device such as a solid state drive (Sold State Drive; SSD), or as a random access memory (Random Access Memory; RAM) used to store temporary necessary information (parameters, arrays, etc.) related to program operations. etc. memory to implement. In addition, a combination of these can also be used. In particular, the storage unit 32 stores various programs related to the information processing device 3 executed by the control unit 33.

In addition, the memory unit 32 also memorizes a learned model that learns the feature quantity f(N) of the desired period (Epoch), the feature quantity f(N±δ) of the near past period, and the correlation with the user's sleep and wakefulness. . Such machine learning algorithms are preferably based on conventional algorithms. For example, logistic regression (Logistic regression), Random forest (Random forest), XGBoost (eXtreme Gradient Boosting, Extreme Gradient Boosting), Multilayer perceptron (MLP), etc. can be cited. In addition, every time the information processing device 3 is used, machine learning using it as teacher data can be further implemented and updated. New previously learned model.

The control unit 33 (an example of a processor) performs processing and control related to the overall operation of the information processing device 3 . The control unit 33 is, for example, a central processing unit (Central Processing Unit; CPU) not shown. The control unit 33 realizes various functions related to the information processing device 3 by reading the predetermined computer program stored in the memory unit 32 . That is, the information processing performed by the software (stored in the memory unit 32) is specifically implemented by the hardware (the control unit 33), and can be executed as each functional unit described later. In addition, in FIG. 2 , although the control unit 33 is shown as a single unit, the present invention is not limited to this and may be implemented with a plurality of control units 33 according to each function. In addition, a combination of these can also be used.

2. Functional structure

In this section, the functional structure of this implementation aspect will be described. As described above, the information processing of the software stored in the memory unit 32 is specifically implemented by the control unit 33 as an example of hardware, and can be executed as each functional unit included in the control unit 33 . FIG. 4 is a functional block diagram of the control unit 33 in the information processing device 3 . That is, the control unit 33 includes the following components.

The acquisition unit 331 is configured to acquire various information from the outside through a network, a storage medium M, or the like. For example, the acquisition unit 331 may acquire a signal indicating the acceleration of at least a part of the user's body. This will be discussed in further detail later.

The band limiting unit 332 is configured to perform band limiting processing on the signal acquired by the acquiring unit 331 . For example, the frequency band limiting unit 332 may limit the signal to a specific frequency component C among the frequency components included in the signal indicating acceleration. This will be discussed in further detail later.

The converting unit 333 is configured to convert the specific frequency component C limited by the band limiting unit 332 The signal is processed by Fourier transformation. For example, a signal composed of a specific frequency component C may be Fourier transformed to generate data for determination. This will be discussed in further detail later.

The correction unit 334 is configured to execute correction processing. For example, the correction unit 334 may standardize and correct the distribution of frequency components of the determination data generated by the conversion unit 333. This will be discussed in further detail later.

The determination unit 335 is configured to determine whether the user is asleep or awake. For example, the determination unit 335 may determine the user's sleep and wakefulness based on the specific frequency component C and preset reference information IF. This will be discussed in further detail later.

The display control unit 336 is configured to generate various display information and control display content visible to the user. Display information can be information generated in a form visible to users such as screens, images, icons, text, etc., or it can also be rendering information (Rendering information) used to display screens, images, icons, text, etc.

3.Information processing methods

In this section, the information processing method of the aforementioned sleep-wake determination system 1 will be explained.

(Process Overview)

FIG. 5 is an activity diagram showing the information processing flow of the sleep-awake determination system 1 . The information processing flow will be outlined below according to each activity in Figure 5. The user here is assumed to wish to use the service provided by the sleep and wakefulness determination system 1 to determine sleep and wakefulness. It should be noted that the determination of sleep and awakening here may include determining whether the user is in a sleeping state or an awakening state. The user wears the wearable device 2 on, for example, his or her arm.

First, the user lies down and rests while wearing the wearable device 2 on at least a part of his or her body, such as one arm. During this period, the acceleration sensor 24 in the wearable device 2 sequentially detects and measures the acceleration of the user's single arm (activity A101). The log data showing the signal of the measured acceleration is sequentially memorized in the memory medium M inserted into the wearable device 2 .

Subsequently, after obtaining sufficient log data of the signal representing the acceleration, the signal representing the acceleration memorized in the memory medium M is transferred to the information processing device 3 by replacing the memory medium M from the wearable device 2 with the information processing device 3 . That is, the acquisition unit 331 acquires a signal indicating the acceleration of at least a part of the user's body (activity A102). The acquired signal indicating the acceleration can be read from the temporary memory area of the memory unit 32 .

Next, the control unit 33 reads a predetermined computer program stored in the memory unit 32 to limit the frequency component included in the acceleration signal to the specific frequency component C. That is, the band limiting unit 332 limits the frequency component included in the signal indicating acceleration to the specific frequency component C, and it is particularly preferable that the band limiting unit 332 limits the frequency component to the specific frequency component C using the high-pass filter HF (Activity A103). According to this aspect, the specific frequency component C can be limited to a high-frequency component above the cutoff frequency, so that the user's sleep and wakefulness can be determined more robustly.

Then, the control unit 33 reads a predetermined computer program stored in the memory unit 32, and Fourier transform is performed on the signal composed of the specific frequency component C. Such Fourier transformation preferably uses the algorithm of FFT (Fast Fourier Transformation; Fast Fourier Transform). That is, the conversion unit 333 performs Fourier transformation on the signal composed of the specific frequency component C and generates data for determination (Activity A104).

Subsequently, the control unit 33 reads a predetermined computer program stored in the memory unit 32, thereby performing standardization of the frequency components of the generated determination data. That is, the correction unit 334 normalizes and corrects the frequency component of the determination data (the distribution of the specific frequency component C) (Activity A105).

Next, the standardized judgment data is processed as a period E defined in units of predetermined time t. Specifically, the control unit 33 reads a predetermined computer program stored in the memory unit 32 to calculate the characteristic amount f(N) for each time period E (activity A106). N here is the sequence number of period E, which will be further detailed later.

After that, the control unit 33 reads the reference information IF stored in the storage unit 32 (Activity A107), and supplies the feature quantity f of the desired time period E and the feature quantity f of the predetermined time period E around it to the reference information IF. This displays the judgment results about the user's sleep and wakefulness. Particularly preferred is a learned model IF1 that has learned the correlation between the specific frequency component C and sleep and wakefulness with reference to the information IF system. According to this aspect, machine learning can be used to determine the user's sleep and wakefulness with high accuracy.

That is, the determination unit 335 determines the user's sleep and wakefulness based on the determination data including the specific frequency component C and the preset reference information IF. Preferably, the determination unit 335 specifies the feature amount f for each period E defined by the predetermined time t based on the specific frequency component C. In addition, the determination unit 335 is based on the feature amount f of the expected time period E among the time periods E, the feature amount f of the near past time period E that is earlier than the expected time period E in the time series, and the learned model IF1 as an example of the reference information IF. , to determine the user's sleep and wakefulness. Then, the display control unit 336 controls the determination result to be displayed in a manner that can be presented to the user (Activity A108). According to this aspect, the user's sleep and wakefulness can be determined with higher accuracy.

To sum up, the aforementioned sleep awakening determination method includes each step in the sleep awakening determination system 1 . In the acquisition step, a signal indicating the acceleration of at least a part of the user's body is acquired. In the band limiting step, the frequency component contained in the signal indicating acceleration is limited to the specific frequency component C. In the transformation step, the signal consisting of the specific frequency component C is Fourier transformed to generate data for determination. In the determination step, the user's sleep and wakefulness are determined based on the determination data and preset reference information.

According to this aspect, sleep and wakefulness can be determined with sufficiently high accuracy using fewer wearing devices. In particular, by first performing frequency band limitation and then performing Fourier transformation, the feature quantity f in the determination data can be made prominent, and as a result, it is helpful to determine sleep and wakefulness. In addition, since it does not depend on the sampling frequency of the signal obtained by the device, the user's sleep and awakening can be determined more robustly.

(Details of information processing)

Here, the details of the information processing outlined in Figure 5 will be explained. 6 and 7 are conceptual diagrams for explaining period E. FIG. 8 shows a graph showing the L2 norm of acceleration for each period (FIG. 8A) and a graph showing a hypnogram (FIG. 8B) of actual sleep. FIG. 9 is a graph showing the accuracy of the learned model IF1 at each sampling frequency. FIG. 10 is a diagram showing the accuracy of sleep/awake determination at each bit rate in analog-to-digital conversion. FIG. 11 is a schematic diagram showing the standardization process performed by the correction unit 334.

As shown in FIG. 6, the period E is the signal log data of the acceleration specified by the predetermined time t. The aforementioned predetermined time t is, for example, 5 seconds to 600 seconds, preferably 10 seconds to 120 seconds, and further preferably 20 seconds to 60 seconds, specifically, for example, 10, 20, 30, 40, 50, 60, 70 ,80,90,100,110,120,130,140,150,160,170,180,190,200,210,220,230,240,250,260, 270, 280, 290, 300, 310, 320, 330, 340, 350, 360, 370, 380, 390, 400, 410, 420, 430, 440, 450, 460, 470, 480, 490, 500, 510, 520, 530, 540, 550, 560, 570, 580, 590, 600 seconds, or within the range between any two of the values shown above.

In this embodiment, the signal indicating acceleration is processed as a plurality of periods E. Suppose that the Nth period E in the time series is called the expected period E, and the N-1 to N-4th period E that is slightly earlier than this is called the near past period E. In this embodiment, the feature quantity f(N) of the desired time period E and the feature quantity f(N-δ) of the recent past time period E are used as inputs, and the determination is made based on the aforementioned learned model stored in the memory unit 32 The user's sleep and wakefulness.

Specifically, the δ value is, for example, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 ,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46 , 47, 48, 49, 50, or within the range between any two of the values shown above. Of course, the numerical value of the period is only an example and is not limited thereto. In addition, as shown in FIG. 7 , the feature quantity f(N±δ) may be used instead of the feature quantity f(N-δ). In this case, for example, the N-4th to N+4th periods E may also be called peripheral periods E.

In addition, the feature value f of the period E is not particularly limited. For example, although the three-dimensional acceleration vector v(x, y, z) indicating acceleration is subjected to high-pass filter HF, FFT, and normalization processing, the feature amount f(N) can also be extracted from the L2 norm and used for Determination of sleep and wakefulness. Specifically, for example, the feature quantity f(N) may be a histogram generated by classifying a scalar value such as L2 norm or its logarithm by a plurality of threshold values, or may be based on a scalar value multiplied by a window function. The power spectrum of the product. For example, Figure 8A shows Logarithmic power spectrum showing the time difference of acceleration. Figure 8B shows a sleep progression diagram related to Figure 8A.

Next, as shown in FIG. 9 , it can be seen that when generating the learned model IF1, the accuracy of the judgment changes depending on the sampling frequency of the teacher material. Specifically, according to the sequence of Figure 8A to Figure 8D , the sampling frequency of the teacher's material is set to 50Hz, 25Hz, 10Hz and 5Hz to generate the learned model IF1.

Overall, when the sampling frequency of the teacher data matches the sampling frequency of the input sample signal, the accuracy score (Score) shows an upward trend. Furthermore, especially when the sampling frequency of the teacher data is 25 Hz or more when the learned model IF1 is generated, higher accuracy can be achieved even if the sampling frequency of the input sample signal is low. When changing the sampling frequency, there is no big difference between the case where resampling processing is performed and the case where thinning processing is performed.

In summary, it is preferable that the learned model IF1 is a learned model IF1 that learns the correlation between the specific frequency component C and sleep and wakefulness by acquiring a signal with a sampling frequency of 5 Hz or more. In addition, the sampling frequency of the sample signal input to the learned model IF1 is preferably 5 Hz or more. For such a learned model, when the sampling frequency of input samples matches the sampling frequency during learning, high-precision determination can be achieved.

Particularly preferably, the learned model IF1 is a learned model IF1 that learns the correlation between the specific frequency component C and sleep and wakefulness by acquiring a signal with a sampling frequency of 25 Hz or more. According to this aspect, even if the sampling frequency of the input sample is lower than the sampling frequency during learning, high-precision determination can be achieved.

That is, the sampling frequency of the signal used in learning the learned model IF1 is specifically, for example, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100Hz, or within the range between any two of the values shown above.

In addition, as shown in Figure 10, the accuracy of the determination of the influence of the bit rate during analog-to-digital conversion was confirmed. As can be seen from Figure 10, in the case of resampling, a difference can be seen between 4 bits and 6 bits, and in the case of thinning, a difference can be seen between 6 bits and 8 bits . That is, it is preferable that the acquisition unit 331 sets the bit rate to 8 bits or more, performs analog-to-digital conversion of the acceleration, and acquires the signal. Specifically, it may be, for example, 8, 10, 12, 14, 16, 32, 48, or 64 bits, or it may be in the range between any two of the numerical values shown here. According to this aspect, the analog-to-digital converted signal can be maintained at a higher accuracy, so that the user's sleep and wakefulness can be determined more robustly. The analog-to-digital conversion itself can be performed with more than 8 bits, or it can be performed in such a way that the obtained signal result is more than 8 bits.

Furthermore, as shown in FIG. 11 , it was confirmed that the frequency sensitivity of each product of the wearable device 2 is different. Here, the case of the wearable device 2a and the wearable device 2b is illustrated. In this embodiment, in order to alleviate such errors caused by the dispersion of frequency sensitivity of each product, a normalization process performed by the correction unit 334 is introduced. By performing the standardization process, both the wearable device 2a and the wearable device 2b change into a substantially uniform frequency sensitivity distribution. That is, in this embodiment, by including the standardization process performed by the correction unit 334, the influence of frequency sensitivity can be reduced regardless of the product, and the user's sleep and awakening can be determined more robustly.

4.Others

The sleep/awakening determination system 1 of this embodiment may also adopt the following aspects.

Although the structure of the sleep and awakening determination system 1 has been described in the above embodiments, a computer program for executing each step in the sleep and awakening determining system 1 may also be provided to at least one computer. According to this aspect, sleep and wakefulness can be determined with sufficiently high accuracy using fewer wearing devices. In addition, since it does not depend on the sampling frequency of the device acquisition signal, the user's sleep and awakening can be determined more robustly.

The log data of the signal indicating acceleration from the wearable device 2 to the information processing device 3 may also be transferred not via the memory medium M, but via a communication network such as the Internet, an intranet, or a dedicated wireless communication. In addition, the measurement of acceleration by the wearable device 2 and the determination of sleep and wakefulness by the information processing device 3 can also be performed online in a somewhat real-time manner with a certain delay.

In this embodiment, the acquisition unit 331 , the band limiting unit 332 , the conversion unit 333 , the correction unit 334 , the determination unit 335 and the display control unit 336 are explained as functional units implemented by the control unit 33 of the information processing device 3 , however, at least a part thereof may also be implemented as a functional unit implemented by the control unit 23 of the wearable device 2 .

In addition, the wearable device 2 and the information processing device 3 may be integrated. That is, the information processing device 3 may be a wearable device 2 that the user can wear on a part of the body, and may further include an acceleration sensor 24. The acceleration sensor 24 may be configured as a three-dimensional acceleration vector v ( x, y, z).

The frequency band limiting unit 332 may be configured to use a band-pass filter instead of the high-pass filter HF to limit the frequency component to the specific frequency component C. According to this aspect, specific frequency components can be limited to optimal components, This enables a more robust determination of the user's sleep and wakefulness.

The order of activities A103 and A104 shown in Figure 5 can also be reversed. That is, the correction unit 334 may normalize and correct the frequency component distribution of the acquired signal, and the band limiting unit 332 may limit the normalized frequency component to the specific frequency component C. According to this aspect, the influence of different frequency sensitivities on each device can be reduced, so that the user's sleep and wakefulness can be determined more robustly.

In addition to the acceleration sensor 24 of this embodiment, other sensors may be added as appropriate. For example, you can add a SpO2 sensor, ambient light sensor, heart rate sensor, etc. It can also be implemented as a SpO2 sensor that can measure transcutaneous arterial blood oxygen saturation, and the results are additionally used to determine sleep and wakefulness. It can also be implemented as an ambient light sensor that can measure the user's ambient light intensity, and the results are additionally used to determine sleep and wakefulness. It can also be implemented as a heart rate sensor that can measure the user's heart rate, and the results are additionally used to determine sleep and wakefulness.

In addition, various aspects described below can also be provided.

(1) A sleep awakening determination system, which includes at least one processor. The at least one processor can execute a computer program to perform the following steps: in the obtaining step, obtain a signal indicating the acceleration of at least a part of the user's body; In the band limiting step, the frequency component included in the signal indicating the acceleration is limited to a specific frequency component; in the converting step, the signal consisting of the specific frequency component is Fourier transformed to generate determination data; in the determining step, Based on the aforementioned determination data and preset reference information, the user's sleep and awakening are determined.

(2) The sleep awakening determination system according to the above (1), wherein in the frequency band limiting step, a high-pass filter or a band-pass filter is used to limit the frequency component to the specific frequency component.

(3) The sleep/awakening determination system according to (1) or (2) above, wherein in the correction step, the distribution of frequency components of the determination data is normalized and corrected.

(4) The sleep and awakening determination system according to the above (1) or (2), wherein in the correction step, the obtained distribution of the frequency components of the signal is normalized and corrected, and in the frequency band limiting step, The normalized frequency component is limited to the specific frequency component.

(5) The sleep awakening determination system according to any one of the above (1) to (4), wherein in the above acquisition step, the bit rate is set to 8 bits or more, and the above acceleration is converted into analog to digital, Obtain the aforementioned signal.

(6) The sleep and wakefulness determination system according to any one of the above (1) to (5), wherein the reference information is a learned model that has learned the correlation between the specific frequency component and the sleep and wakefulness.

(7) The sleep wakefulness determination system according to the above (6), wherein the learned model is a learned model that learns the correlation between the specific frequency component and the sleep and wakefulness by acquiring the signal at a sampling frequency of 5 Hz or more.

(8) The sleep and wakefulness determination system according to the above (7), wherein the learned model is a learned model that learns the correlation between the specific frequency component and the sleep and wakefulness by acquiring the signal at a sampling frequency of 25 Hz or more.

(9) The sleep and awakening determination system according to any one of the above (1) to (8), wherein in the determination step, the characteristic amount is determined for each period defined by a predetermined time based on the determination data. , based on the aforementioned feature quantity of the desired period in the aforementioned period, is smaller in time series than the aforementioned desired period The aforementioned feature quantity and the aforementioned reference information in the earlier recent past period are used to determine the sleep and wakefulness of the aforementioned user.

(10) A method for determining sleep arousal, which includes each step in the sleep arousal determining system described in any one of the above (1) to (9).

(11) A computer program that causes at least one computer to execute each step in the sleep-awakening determination system described in any one of (1) to (9) above.

Of course, it doesn't stop there.

Finally, although the present invention has been described above in various embodiments, these are only provided as examples and are not intended to limit the scope of the present invention. This novel embodiment can be implemented in various other forms, and various omissions, substitutions and changes can be made without departing from the spirit of the invention. This embodiment and its modifications are included in the scope and gist of the invention, and are included in the scope of the invention described in the patent application and its equivalent scope.

1: Sleep and wakefulness determination system 2: Wearable devices 3:Information processing device

Claims (11)

A sleep awakening determination system, which includes: at least one processor. The at least one processor can execute a computer program to perform the following steps. In the obtaining step, a signal representing the acceleration of at least a part of the user's body is obtained. In the frequency band limit, In the step, the frequency component included in the signal indicating the acceleration is limited to a specific frequency component. In the conversion step, the signal composed of the specific frequency component is Fourier transformed to generate data for determination. In the determination step, based on the above The determination data and preset reference information are used to determine the sleep and wakefulness of the aforementioned user. The sleep and awakening determination system according to claim 1, wherein in the frequency band limiting step, a high-pass filter or a band-pass filter is used to limit the frequency component to the specific frequency component. The sleep and awakening determination system according to claim 1, wherein in the correction step, the distribution of the frequency components of the determination data is standardized and corrected. The sleep awakening determination system according to Claim 1, wherein in the correction step, the distribution of the frequency components of the obtained signal is standardized and corrected, and in the frequency band limiting step, the standardized frequency components are limited is the aforementioned specific frequency component. The sleep and wakefulness determination system according to claim 1, wherein in the aforementioned acquisition step, the bit rate is set to 8 bits or more, and the aforementioned acceleration is subjected to analog-digital conversion to obtain the aforementioned signal. The sleep and wakefulness determination system according to claim 1, wherein the reference information is a learned model that has learned the correlation between the specific frequency component and the sleep and wakefulness. The sleep and wakefulness determination system according to claim 6, wherein the learned model is a learned model that learns the correlation between the specific frequency component and the sleep and wakefulness by acquiring the signal at a sampling frequency of 5 Hz or more. The sleep and wakefulness determination system according to claim 7, wherein the learned model is a learned model that learns the correlation between the specific frequency component and the sleep and wakefulness by acquiring the signal at a sampling frequency of 25 Hz or more. The sleep and awakening determination system according to claim 1, wherein in the determination step, the characteristic amount is determined for each period (Epoch) defined by a predetermined time based on the determination data, and the characteristic amount is determined based on the desired period in the aforementioned period. The aforementioned feature value, the aforementioned feature value in a past period that is earlier than the aforementioned desired time period in time series, and the aforementioned reference information are used to determine the sleep and wakefulness of the aforementioned user. A sleep awakening determination method, which includes each step in the sleep awakening determination system described in any one of claims 1 to 9. A computer program that causes at least one computer to execute each step in the sleep-awakening determination system described in any one of claims 1 to 9.